The concept of producing important pharmaceutical and nutriceutical proteins in transgenic animals is now firmly established (Van Cott, K. E. and Velander, W. H., Exp. Opin. Invest. Drugs, 7 (10): 1683-1690 (1998)), with three potential products, alpha-1 antitrypsin, antithrombin III and alpha glucosidase in the late stages of clinical trials. These proteins, and nearly all other transgenic polypeptides being developed commercially, were produced from a single DNA construct designed to produce a single polypeptide. In general terms, this “classical” design incorporates three distinct regions of DNA, which are all joined or operably linked in one contiguous strand.
The first region of DNA is a tissue specific promoter, in the above mentioned examples a milk protein promoter, which directs expression of the gene to a target organ, the mammary gland, which is regulated by lactogenic hormones, growth factors, cell-cell and cell-substratum interactions. The second region of DNA is the coding region, which may consist of complimentary DNA (cDNA, containing no introns), genomic DNA (gDNA) or a combination of both in a format called a mini-gene. It is important to note that cDNAs, and perhaps also minigenes, have a silencing effect (failure to express or poor expression levels) on adjacent transgenes (Clark, A. J., et al., NAR, 25 (5), 1009-1014, 1997). Therefore, a method of overcoming this silencing effect using non-genomic DNA sequences is highly desirable. The coding region contains the information needed to produce a specific protein, including any processing and secretory signals. The third region, the 3′ region, contains further regulatory sequences and may influence the quantity of polypeptide that is produced from that construct. Non-genomic DNA sequences are inherently smaller than gDNA sequences and are therefore, much easier to manipulate in classical transgene formats.
Although this classical design has been successful in producing commercially viable quantities of certain proteins, there are two areas in which this system is not optimal. First, it is generally accepted that using cDNAs or minigenes in a classically designed construct, is less efficient for protein production than using a corresponding gDNA coding region. Indeed, this is such a problem that methods have been developed to address this issue (Clark, A. J. et al, Biotechnology 10, 1450-1454, 1992). Whilst these methods can improve the efficiency and level of expression of cDNAs and minigenes to some extent, they do not improve expression to the same level as is typically obtained using gDNA. A higher level would be ideal for commercial protein production.
The second area in which the classical single gene DNA construct design is suboptimal is in the production of highly biologically active proteins in transgenic animals. Proteins with an extremely high biological activity can be detrimental to the transgenic animal, even if circulatory levels (or other systemic levels) are low (Castro, F. O., et al., Selection of Genes for Expression in Milk: The Case of the Human EPO Gene, in Mammary Gland Transgenesis. Therapeutic Protein Production. Castro and Janne (eds.) Springer-Verlag Berlin New York, 91-106, 1998). This can be due in large part to either ectopic expression (expression of the transgene in organs other than the targeted one) or leakage of the protein product into the blood from the target organ. If the protein product is highly biologically active, expression ideally must be strictly controlled so that the animal is exposed to the product for a short time only, thus reducing the chance of any lasting detrimental effects. This requires an expression system that can be turned on and off very rapidly and precisely.
Regulation of Promoters
The expression of many genes is controlled at the level of transcription, when the DNA sequences are transcribed into RNA, prior to being translated into protein (Latchman, D. S., Eukaryotic Transcription Factors, Academic Press, 1998). The DNA sequence element that controls transcription is the promoter. This generally contains a small core region, which is capable of directing constitutive or basal levels of transcription, and upstream response elements that control spatial and temporal regulation of transcription. These DNA sequences include two types of elements, those which are involved in the basic process of transcription and are found in many genes exhibiting distinct patterns of regulation, and those found only in genes transcribed in a particular tissue or in response to a specific signal. The latter elements likely produce this specific expression pattern. They are binding sites for a wide range of different cellular proteins (transcription factors) whose levels fluctuate in response to stimuli from external or internal sources. Gene expression in a given tissue may be stimulated or inhibited depending on the type and amount of transcription factors that are present in that tissue at any time. Many transcription factors or other proteins that enable transcription factor pathways are largely uncharacterized from the perspective of an exact biochemical analysis, which details their conformationally-dependent interactions with DNA. Overall, the regulation of expression at the DNA level, is a function of which regulatory elements (binding sites) are present in the promoter and how the cell or tissue responds to its environment by changing the relative levels of the different DNA binding transcription factors in the cell.
Another mechanism involved in the precise control of gene expression is transcriptional repression (Maldonado, E., et al, Cell, 99 (5), 455458, 1999). Transcriptional repressor proteins associate with their target genes either directly through a DNA-binding domain or indirectly by interacting with other DNA-bound proteins. The repressor protein can inhibit transcription by masking a transcriptional activation domain, blocking the interaction of an activator with other transcription components or by displacing an activator from the DNA.
Milk protein genes are characterized by a strict tissue specific expression and regulation during the process of functional differentiation. They are coordinately expressed in response to various developmental signals, such as changing levels of lactogenic hormones prolactin, insulin, glucocorticoids, progesterone), local levels of certain growth factors (EGF), cell-cell interactions and interactions with extra-cellular matrix (ECM) components (Rijnkels, M. and Pieper, F. R., Casein Gene-Based Mammary Gland-Specific Transgene Expression, in Mammary Gland Transgenesis. Therapeutic Protein Production. Castro and Janne (eds.) Springer-Verlag, Berlin, New York, 41-64, 1998).
Lactogenic hormones activate latent transcription factors in the cytoplasm of mammary epithelial cells. The steroid hormones progesterone, estrogen, and glucocorticoid regulate the transcription of target genes by binding to specific intracellular receptors. Some models purport that binding of the hormone with its receptor changes the receptor's conformation from a physiologically inactive form to a form which is active and capable of dimerization. The active receptors are then capable of binding specific DNA sites in the regulatory region of the target gene promoters, stimulating gene transcription and thus, protein synthesis. Steroid receptors belong to a superfamily of ligand-inducible transcription factors and it has been well documented that these are modular proteins organized into structurally and functionally defined domains. It has also been shown that these domains can be rearranged as independent cassettes within their own molecules or as hybrid molecules with domains from other regulatory peptides. Interestingly, the transactivation domains of the glucocorticoid receptor can be duplicated in tandem and show positional independence in a “super receptor” with 3-4 times the activity of the wild type protein. (Hollenberg, S. M. and Evans, R. M., Cell, 55, 899-906, 1988; Fuller, P. J., FASEB J., 5, 3092-3099, 1991; U.S. Pat. No. 5,364,791; U.S. Pat. No. 5,935,934; Whitfield, G. K., et al, J. Cell. Biochem., suppl. 32-33, 110-122, 1999; Braselmann, S., et al, PNAS, 90, 1657-1666, 1993). The structure and function of the steroid receptor superfamily is well conserved. Generally there are three main domains and several sub-domains or regions. The NH2-terminal domain is the least conserved in size and sequence and contains one of the two, transactivation sequences of the receptor. The central DNA binding domain of about 70 amino acids is highly conserved, as is the COOH-terminal ligand binding domain. This latter domain also contains sub-domains responsible for dimerization, heat shock protein (hsp) 90 binding, nuclear localization and transactivation.
Prolactin plays the essential role in milk protein gene expression and exerts its effect through binding to the extracellular domain of the prolactin receptor and through receptor dimerization. This activates a protein tyrosine kinase (JAK2) which is non-covalently associated with the cytoplasmic domain of the prolactin receptor (Gouilleux, F., et al, EMBO J., 13 (18), 4361-4369, 1994; Imada, K and Leonard, W. J., Mol. Immunol., 37 (1-2), 1-11, 2000). The activated JAK2 phosphorylates the signal transducer and transcription activator, Stat 5, causing it to dimerize and subsequently, translocate to the nucleus. Once in the nucleus, Stat5 specifically binds to sequence elements in the promoter regions of milk protein genes (Liu, X., et al, PNAS, 92, 8831-8835, 1995; Cella, N., et al, Mol. Cell. Biol., 18 (4), 1783-1792, 1998; Mayr, S., et al, Eur. J. Biochem., 258 (2), 784-793, 1998). In an analysis of 28 milk protein gene promoters (Malewski, T., BioSystems, 45, 29-44, 1998) there were 4 transcription factor binding sites that were present in every promoter, C/EBP, CTF/NF1, MAF and MGF (Stat 5). Although steroid hormone receptors and Stat factors comprise two distinct families of inducible transcription factors their basic structure is similar. Stat proteins are modular with an amino terminus that regulates nuclear translocation and mediates the interaction between Stat dimers (Callus, B. A. and Mathey-Prevot, B., J. Biol. Chem., 275 (22), 16954-16962, 2000). There is a central DNA binding domain and a carboxy terminal region, which contains the phosphorylation site and a transactivation domain.
Egg white genes seem to be regulated in a similarly complex manner. It is known that the progesterone-dependent activation of the egg white genes in the chicken oviduct is mediated through the progesterone receptor (Dobson, A. D. W., et al, J. Biol. Chem., 264 (7), 4207-4211, 1989). In addition, the chicken ovalbumin upstream promoter-transcription factor (COUP-TF) is a high affinity and specific DNA binding protein, which interacts as a dimer with the distal promoter sequence of the ovalbumin gene and promotes initiation of transcription of this gene by RNA polymerase (O'Malley, B. W. and Tsai, M-J., Biol. Reprod., 46, 163-167, 1992). COUP-TFs are orphan members (no binding ligand has as yet been determined for these receptors) of the nuclear receptor superfamily, and have been shown to play a key role in the regulation of organogenesis, neurogenesis, metabolic enzyme production and cellular differentiation during embryogenic development, via transcriptional repression and activation (Sugiyama, T., et al, J. Biol. Chem., 5 (5), 3446-3454, 2000).
A protein expression method based on the inducible Tet repressor system has been developed (Furth, P. A., et al, PNAS, 91, 9302-9306, 1994), but the levels of basal expression without induction are too high to be useful in transgenic animals (Soulier S. et al, Eur. J. Biochem. 260, 533-539, 1999). Another inducible system based on the use of the ecdysone receptor has been reported (No, D., et al, PNAS, 93, 3346-3351, 1996; PCT 97/38117, PCT 99158155) and has recently given encouraging results in transgenic mice (Albanese, C., et al, FASEB J., 14, 877-884, 2000). However, this system required the delivery of an exogenous ligand to the mice for the full lactation period. Such a ligand would be costly and difficult to procure for regular administration in a production environment.
A new multi-gene system for protein production in transgenic animals would improve commercial levels of production from cDNA constructs by amplifying specifically tailored transcription factors which need not naturally occur in the tissue targeted for expression, but would be transgenically expressed specifically in that tissue. Unlike classical gene expression formats for recombinant proteins, the tissue specific promoter would not be linked to the protein to be expressed, but would be used to drive expression of transcription factors which do not have a signal sequence and so are not secreted. In addition, the added control that a doubly inducible multi-gene system would provide, which is inexpensive and easily applied, could enable the production of highly biologically active proteins in transgenic animals in a pulsatile fashion so as to avoid longterm detrimental effects.
Proteins for Transgenic Production
A multi-gene system, as described below, can be used to direct expression of any protein, particularly any secreted protein, which can be expressed in a transgenic organism in useful quantities, either for research or commercial development. Particular proteins of interest with respect to production by multi-gene expression systems include relaxin and other hormones with cross-species activity such as growth factors, erythropoietin (EPO) and other blood cell growth stimulating factors. For these proteins, the expression may be problematic in terms of harming the host animal as is known to happen when EPO is expressed for an extended period of time. It is noted that tissue specific expression of transgenes is not an absolute phenomenon and promiscuous expression or systemic transport of the expressed recombinant protein within the animal almost always occurs with any expression system in any animal, albeit at very low levels. However, even at low levels of expression of EPO, when the EPO is expressed over an extended period of time, the hematocrit of the host animal can rise to a fatal level. Thus a temporal control which can enable pulse expression using an external inducer molecule could overcome the problems of continuous and extended expression (ie., as could occur if expression occurs over an entire lactation period). Pulse or truncated expression would be useful in preventing an adverse, systemic physiologic effect by recombinant molecules like EPO, which can cause these effects at very low levels.
Relaxin is widely known as a hormone of pregnancy and parturition and typically circulates at less than 50 pg/ml in the blood of women. However, it is now emerging that the peptide has a far wider biological function than was at first thought. There are receptor sites for relaxin in striated muscle, smooth muscle, cardiac muscle, connective tissue, the autonomic and the central nervous systems. Human relaxin has been demonstrated to inhibit excessive connective tissue build-up and is in Phase II trials for the treatment of Scleroderma Porcine relaxin was available commercially in the 1950-60s and was used extensively for such conditions as cervical ripening, scleroderma, premature labour, PMS, decubital ulcers and glaucoma. Relaxin is known to adversely affect the lactation of different mammalian species but does not seem to affect the pig in a similar manner. Therefore, the pig is perfectly suited for production of relaxin in milk.
Other examples of proteins which it would be desirable to produce in transgenic organisms, are proteins that are protease inhibitors. Some examples of protease inhibitors are Alpha 1-antitrypsin, Alpha 2 Macroglobulin, and serum leukocyte protease inhibitor. These proteins are serine protease inhibitors that show antiviral, non-steroidal anti-inflammatory and wound healing properties. These proteins are useful in veterinary, cosmetic and nutriceutical applications.
Alpha 1-antitrypsin (AAT) is a naturally occurring glycoprotein produced by the liver. Improperly glycosylated recombinant AAT such as made by yeast, does not have a sufficient circulation half-life to be used as a parenterally administered therapeutic. Congenital deficiency results in the condition emphysema and in 1985 Bayer Pharmaceuticals began marketing a plasma derived AAT product, Prolastin. Unfortunately, due to shortages of Asafe@ plasma and frequent recalls, supplies of Prolastin are often very limited. AAT has also been used to treat psoriasis, atopic dermatitis, ear inflammation, cystic fibrosis and emphysema, and to assist in wound healing. It has been estimated that over 10 million people in the US alone may benefit from AAT therapies.
Alpha 2 macroglobulin (A2M) is a very large, complex glycoprotein with a published cDNA sequence containing 1451 amino acids. The mature protein is a tetrameric molecule composed of four 180 kDa subunits and thus has a molecular weight which is over 720 kDa. Its complexity makes it most suited for production in mammalian systems but few mammalian systems will likely make A2M at commercially viable levels. A2M is indicated for treatment of asthma, bronchial inflammation and eczema and acts as a protease inhibitor to both endogenous and exogenous proteases that cause inflammation. A2M is necessarily more potent than alpha 1-antitrypsin due to its irreversible binding of target proteases. A2M is also useful in inhibiting proteases frequently found in (thermal) burn wounds from yeast and other infections. The high specific activity of these types of proteases allows for smaller doses during treatment. Thus, A2M=s complexity and specific activity make it ideally suited for production in transgenic pig mammary glands.
Vitamin K-Dependent Proteins
Vitamin K-dependent (VKD) proteins such as those proteins associated with haemostasis have complex functions which are largely directed by their primary amino acid structure. In particular, the post-translational modification of glutamic acids in the amino terminal portion of these molecules is essential for proper biological activity. This includes biological activity of both pro-coagulation and anti-coagulation. This particular domain found in VKD-proteins is called the “gla domain”. For example, the Gla domain is an essential recognition sequence in tissue factor (TF) mediated pro-coagulation pathways. The anti-coagulation of this pathway depends upon the lipoprotein-associated coagulation inhibitor, termed LACI, which is a non-VKD protein. LACI forms a complex with the Gla domain of factor Xa, factor VIIa, and TF. Specifically, the Gla domain of factor Xa (FXa) is needed for this procoagulation inhibitory activity. It has been shown that recombinant chimeric molecules having LACI inhibitor (Kunitz type) regions and the Gla domain of FXa can be inhibitory of the TF pathway.
TABLE 1VKD proteins.Protein CFactor X (FX)Bone Gla protein (Osteocalcin)Protein SProthrombinProtein ZFactor VIIFactor IX
Gamma-carboxylation is required for calcium-dependent membrane binding. All of the proteins listed in Table 1 have multiple Gla-residues in a concentrated domain. The Gla-domains of these proteins mediate interaction and the formation of multi-protein coagulation protein complexes. Mammalian coagulation (here collectively meaning both pro-coagulation and anti-coagulation pathways and mechanisms) physiology requires that nearly complete-carboxylation of VKD-proteins occurs within the respective Gla domain for each of these proteins to be maximally functional. Notably, in the context of recombinant synthesis of any protein containing Gla-domains, the extent of gamma-carboxylation of VKD-proteins varies from one mammalian cell source to another, including differences between species and tissue within a species.
VKD-proteins of interest with respect to production by single or multi gene expression systems include those in Table 1, particularly blood clotting factor IX, Protein C and chimeric hybrid vitamin K-dependent proteins. Factor IX is an essential blood clotting protein. Haemophilia B is a genetic disorder in which the production of active Factor IX is defective. It is an inherited disorder that primarily affects males, at the rate of approximately 1 in 30,000. The consequent inability to produce sufficient active Factor IX can lead to profuse bleeding, both internally and externally, either spontaneously or from relatively minor injuries.
In spite of techniques available to amplify recombinant synthesis of VKD proteins such as Protein C and Factor IX, biologically functional recombinant versions of these proteins are difficult to produce and are made typically at levels less than about 0.1 grams per liter per 24 hours in recombinant cell culture media (Grinnell, B. W., et al, in Protein C and Related Anticoagulants. Bruley, D. F. and Drohan, W. N. (eds.), Houston, Tex.; Gulf Publishing Company, 29-63, 1990), or less than 0.22 gm per liter per hour in the milk of transgenic livestock (Van Cott, K. E., et al., Genetic Analysis: Biomolecular Eng., 15, 155-160, 1999). The expression of high levels of FIX using a cDNA construct is difficult. However, the gDNA of FIX, at 33 kbp, is rather large and difficult to manipulate, particularly when compared to the FIX cDNA, which is only 1.4 kbp.
Most VKD-blood plasma proteins are also glycosylated. The extent and types of glycosylation observed is heterogeneous and varies considerably in all species and cell types within a species. Examples of the heterogeneity, structure function relationships of glycosylation are cited by Degen, Seminars in Thrombosis and Hemostasis, 18 (2), 230-242, 1992; Prothrombin and Other Vitamin K Proteins, Vols I and II, Seegers and Walz, Eds., CRC Press, Boca Raton, Fla., 1986.
Glycosylation is a complex post translational modification that occurs on many therapeutic proteins. The process of glycosylation attaches polymeric sugar compounds to the backbone of a protein. These sugar-based structures impart not only an immunologically specific signature upon the protein, but also can change the specific level of activity that the protein has with relation to how long it can reside in the bloodstream of a patient, or how active the protein is in its basic function. All three of these facets can make or break the protein in its role as a therapeutic or wellness product. For example, genetically engineered yeast can impart glycosylation that results in an immunologically adverse signature, which can stimulate the body to make antibodies and essentially reject the protein. In fact, that is part of the reason why yeast vaccines are effective; they easily induce an immune response. The mammary gland of ruminants produces a substantial fraction of glycosylation on milk proteins, which resemble the primitive sugars found in yeast. Thus, applications that result in the long term, repeated exposure of proteins containing yeast or yeast-like signatures, to human tissue are intensely scrutinized with respect to the potential of adverse immune reactions. This structure is also apt to cause dysfunction with respect to the protein=s natural activity and may also contribute to a shortened residence time in blood. In contrast, the mammary gland of pigs gives a glycosylation signature that more closely resembles that found in normal human blood proteins, helping to assure biochemical function and a long circulatory half-life.
The complex post-translational modifications of therapeutic proteins, such as those discussed above that are necessary for physiological activities, pose a difficult obstacle to the production of active vitamin K dependent proteins in cells using cloned genes. Moreover, attempts to culture genetically altered cells to produce VKD polypeptides have produced uneconomically low yields and, generally, preparations of low specific activity. Apparently, the post-translational modification systems in the host cells could not keep pace with production of exogenously encoded protein, reducing specific activity. Therefore, cell culture production methods have not provided the hoped for advantages for producing highly complex proteins reliably and economically.
An attractive alternative is to produce these complex proteins in transgenic organisms. However, it is likely that only mammals and perhaps birds will be able to carry out all the post-translational modifications necessary for their physiological function. It has not been possible, as yet, to produce commercially viable levels of certain complex polypeptides from a controlled source in a highly active form with a good yield, and there exists a need for better methods to produce such proteins.
An interesting new class of proteins, which is likely to be difficult to produce in commercial quantities in cell culture, is the genetically engineered fusion, chimeric and hybrid molecules that are now being developed. These proteins are designed and produced by combining various domains or regions from different natural proteins, either wild type or mutated, which can confer the properties of each domain or region to the final hybrid molecule. An example of this is XLCLACIK1 (Girard, T. J., et al., Science 248, 1421-1424, 1990) which is a hybrid protein made up of domains from factor X and lipoprotein-associated coagulation inhibitor (LACI). LACI appears to inhibit tissue factor (TF)-induced blood coagulation by forming a quaternary inhibitory complex containing FXa, LACI, FVIIa and TF. XLCLACIK1 directly inhibits the activity of the factor VIIa-TF (tissue factor) catalytic complex, but is not dependent on FXa Gamma-carboxylation of the FX portion of the hybrid protein is required for inhibitory activity. In order for efficient carboxylation to occur at high levels, it is likely that the pro-peptide of the recombinant VKD-protein must be properly matched to the endogenous carboxylase system (Stanley, T. B., et al, J. Biol. Chem., 274 (24), 16940-16944, 1999). This is probably true for all VKD-polypeptides including chimeric ones such as XLCLACIK1. It appears that the endogenous carboxylase systems of any given species or tissue within that species, most of which are not identified or characterized, will differ in their compatibility to any given pro-peptide sequence. Also it is frequently desirable to have the pro-peptide cleaved from the nascent VKD protein, such as a XLCLACIK1 polypeptide, once gamma-carboxylation has been completed on the polypeptide's gla domain. It is therefore, also important to find a propeptide sequence that will be efficiently cleaved within the specific species and tissue in which it is being recombinantly produced. These factors render it problematic to find an expression system, which can produce desirable amounts of biologically active VKD-proteins such as XLCLACIK1 chimeric proteins. In spite of being known as a potent coagulation inhibitor since the early 1990s, XLCLACIK1 chimeric molecules have not been made in large amounts in a commercially viable manner (ie., greater than 0.1 gm per liter per 24 hours) in recombinant mammalian cell culture. One way to improve expression of this protein in a transgenic system, particularly in transgenic pigs, may be to substitute the FIX propeptide sequence for the FX propeptide sequence, such a protein would be termed 9XKI.
New therapeutic molecules are being designed to have increased activity, decreased inactivation, increased half-life or specific activity and reduced immunogenicity and/or immunoreactivity to existing circulating antibodies in patients' bloodstreams. This has been demonstrated in genetically engineered Factor VIII proteins (U.S. Pat. No. 5,364,771, U.S. Pat. No. 5,583,209, U.S. Pat. No. 5,888,974, U.S. Pat. No. 5,004,803, U.S. Pat. No. 5,422,260, U.S. Pat. No. 5,451,521, U.S. Pat. No. 5,563,045). Mutations include deletion of the B domain (Lind, P., et al., Eur. J. Biochem. 232, 19-27, 1995), domain substitution or deletion, covalent linkage of domains, site-specific replacement of amino acids and mutation of certain cleavage sites. In particular, a genetically engineered inactivation-resistant factor VIII (IR8) has been developed to help in the treatment of hemophilia A (Pipe, S. W. and Kaufman, R. J., PNAS 94, 11851-11856, 1997). The introduction of specific sequences from porcine factor VIII can also be useful in the formation of a recombinant FVIII which is used to treat hemophiliacs with improved properties as stated above. These molecules can also be designed for improved expression. It is widely known that FVIII has restrictions in intracellular trafficking which lead to low levels of secretion. Modification of the domains associated with intracellular interactions with immunoglobulin binding protein (Bip) or calnexin would be examples of modifications used to improve secretory processing efficiency (Kaufman, R. J., Abstract S1-8, 10th Int. Biotech. Symp., Sydney, Australia, 25-30th Aug., 1996). Factor VIII gDNA is another example of an extremely large and unwieldy DNA sequence (˜110 kbp), whereas the cDNA is only 7 kbp, making it much more manageable.
Whey acidic protein (referred to as “WAP”) is a major whey protein in the milk of mice, rats, rabbits and camels. The regulatory elements of the mouse WAP gene are entered in GenBank (U38816) and cloned WAP gene DNAs are available from the ATCC. The WAP promoter has been used successfully to direct the expression of many different heterologous proteins in transgenic animals for a number a years (EP0264166, Bayna, E. M. and Rosen, J. M., NAR, 18 (10), 2977-2985, 1990). Lubon et al (U.S. Pat. No. 5,831,141) have used a long mouse WAP promoter (up to 4.2 kbp) to produce Protein C in transgenic animals. However, the longest rat WAP promoter that has been used is 949 bp (Dale, T. C., et al., Mol. Cell. Biol., 12 (3), 905-914, 1992).